Apply shear stress by 'fix aveforce'

Thanks again, Eric. Because my model is really small (thousand atoms), it is really hard to use a barostate. The applied stress that I want to apply may be smaller than the stress fluctuation.
Yes, I agree that extracting Cauchy stress tensor is not direct with MD level of small system. So that I want to define a local shear stress by applying force. It is directly from the definition
of stress=force/area. But the problem is the stress defined in this way is larger than my imagination. Small defined stress can drive plastic deformation of crystal.
I don’t know what happens during I use the dimension analysis stress =force (eV/A)/Area (A^2)= eV/A^3= 160.2 GPa. (lammps metal units)

I don’t know if this thread is dead yet, or you figured something out(I am certainly not in this field). But from a non-equillibrium persective, you may want to change your approach. Presumably your system is stable and has some non-equilibrium steady state. One would expect that your system attains a steady state where the force resisting movement equals the force applied. Wouldn’t it make more sense to try and obtain some constant velocity difference to observe the force/stress? This keeps you more in the spirit of Onsager’s reciprocity relationships of either prescribing fluxes or prescribing the force. To the usefulness of this approach, to you, I cannot attest.